Performance of ERA5 Products in Capturing Pluvial and Fluvial Inundation: A Statistical-Cum-Hydrodynamic Framework for Resource-Constrained Large Flood-Prone Watersheds

Flood inundation modelling antecedent to efficient flood management is often constrained by the limited availability of in-situ hydro-meteorological observations. Reanalysis datasets, such as ERA5, offer a promising alternative by providing continuous and high-resolution spatiotemporal information. While ERA5 has been widely recognized for hydrological applications, its performance at the inundation scale remains underexplored. This study rigorously evaluates the performance of ERA5 products for pluvial and fluvial flood inundation over an acute flood-affected watershed in India. Inherent biases of ERA5 rainfalls and GloFAS discharges are realized against station-level observations through probability density and a set of statistical metrics at both median and extreme percentiles (95th and 99th levels). Subsequently, both datasets are forced to LISFLOOD-FP-a global high fidelity flood model, to simulate past flood incidents (2006, 2008, and 2011). Hit rates exceeding 0.8 when comparing simulated inundation with MODIS satellite imagery, establish the suitability of ERA5 for capturing flood inundation. In the later step, design rainfalls and discharges for 50-, 100-, and 200-year recurrence periods were generated using extreme value distributions, enabling the development of high-resolution design flood inundation hazard maps (expressed as flood depth and momentum of flood waves). In the quest for a suitable product to augment data-scarcity challenges, the present study lays a strong foundation on the fidelity of ERA5 for capturing hydrodynamic reality. The design flood maps derived from the study add to the existing database of flood management plans, and provide vital information for refurbishing through appropriate structural and non-structural options.